Single server multi-client message shuffling
The multi-client single-server message shuffling scheme addresses the issue of source identification and security by encrypting and shuffling messages across clients, enhancing data privacy and reducing communication costs, ensuring robustness and efficiency in data transmission.
Patent Information
- Application Number
- PCT/US2024/028049
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2023-05-05
- Filing Date
- 2024-05-06
- Publication Date
- 2024-11-14
AI Technical Summary
In systems where multiple clients transmit data to a single server, the server can identify the source of each message, compromising data security and client privacy, and existing solutions are not robust against resource-constrained or malicious clients, with high per-client communication costs.
Implementing a multi-client single-server message shuffling scheme where clients encrypt and shuffle ciphertexts, obscuring the source identity and ensuring security even with resource-constrained or malicious clients, while reducing per-client communication costs by distributing encryption and shuffling operations across clients.
The scheme effectively obfuscates the source of messages from the server, maintains message security, and reduces per-client communication costs by a significant order of magnitude, making it robust against client dropouts and malicious actors.
Smart Images

Figure US2024028049_14112024_PF_FP_ABST
Abstract
Description
[0001]Attorney Docket No.56113-0414WO1 SINGLE SERVER MULTI-CLIENT MESSAGE SHUFFLING CROSS-REFERENCE TO RELATED APPLICATION This application claims priority to U.S. Provisional Application No.63 / 500,565, filed on May 5, 2023. The disclosure of the prior application is considered part of and is incorporated by reference in the disclosure of this application. BACKGROUND This specification relates to transmitting data between clients and servers. Many real-world applications require multiple clients to each provide data to single server. For example, the server can collect analytics data, e.g., error reports, performance data, or other data, from the clients. The analytics data can be information about the operation of software running on the clients or otherwise being monitored by the clients. As another example, the server can collect training data from the respective clients for use in training or re-training one or more machine learning models. SUMMARY This specification describes a system implemented as computer programs on one or more computers that performs single server, multi-client message shuffling. In other words, this specification describes a system that modifies the communications between a server and a set of multiple clients by causing some or all of the multiple clients to shuffle and re-encrypt ciphertexts generated from messages that are sent by the clients for use from the server. Particular embodiments of the subject matter described in this specification can be implemented so as to realize one or more of the following advantages. Many real-world applications require a server to collect data, i.e., receive messages, from multiple different clients. However, in many of these applications, it is necessary or desirable that the server not be able to determine which message originated from which client and that the contents of each message transmitted by any given client are kept secure, i.e., are not discernible by any of the other clients. This specification describes techniques that, by employing a multi-client single server shuffling scheme, allow the clients to provide messages to the server while obfuscating from the server the identity of the source client for any given message. Attorney Docket No.56113-0414WO1 Generally, there can be a large number of clients, but any given client may be resource-constrained, have limited network connectivity, or both. Thus, computations assigned to a given client may “drop out” (fail) at any given time during the single server, multi-client message shuffling scheme. Therefore, the system implements the scheme to be robust to a certain number of client “drop outs.” That is, even though a certain number of clients drop out while performing the scheme, e.g., due to poor network connectivity or due to resource constraints that prevent the client from performing its assigned operations, the source of the messages of the remaining clients is still obfuscated from the clients. Moreover, given the large and potentially variable number of clients, one or more of the clients may be malicious actors (i) that are attempting to recover unencrypted versions of messages from other clients or (ii) that are cooperating with the server to allow the server to identify the source of some or all of the messages once the messages have been decrypted, or (iii) both. By performing the multi-client single-server message shuffling scheme, the system ensures that (i) the messages are secure until decrypted on the server even though ciphertexts are transmitted to various clients as part of the scheme and (ii) the source of each message is obfuscated from the server even if the server, potentially in collaboration with certain ones of the clients, attempts to discover the source of any given message. Because the scheme requires clients to encrypt and shuffle messages, the scheme requires additional per-client communication costs. However, because of the way in which the scheme distributes the encryption and shuffling operations across the clients, the described techniques improve per-client communication by one or more orders of magnitude with respect to previous (approximate) shuffling protocols. In particular, the per-client communication cost only grows sub-linearly in the number of clients. The details of one or more embodiments of the subject matter of this specification are set forth in the accompanying drawings and the description below. Other features, aspects, and advantages of the subject matter will become apparent from the description, the drawings, and the claims. BRIEF DESCRIPTION OF THE DRAWINGS FIG.1 shows an example message processing system. FIG.2A is a flow diagram of an example process for transmitting messages between clients and a server. Attorney Docket No.56113-0414WO1 FIG.2B is a flow diagram of an example process for performing key generation. FIG.2C is a flow diagram of an example process for performing shared key decryption. FIG.3A is a flow diagram of an example process for performing a shuffling iteration. FIG.3B is a flow diagram of an example process for carrying out a shuffling scheme. FIG.3C is a flow diagram of an example process for carrying out an amortized shuffler scheme. FIG.4A shows an example of an alternating shuffler scheme FIG.4B is a flow diagram of an example process for carrying out an alternating shuffler scheme. Like reference numbers and designations in the various drawings indicate like elements. DETAILED DESCRIPTION FIG.1 shows an example message processing system 100. The message processing system 100 is an example of a system implemented as computer programs on one or more computers in one or more locations, in which the systems, components, and techniques described below can be implemented. In particular, the system 100 includes multiple clients 110 and a server 120. Each client 110 can be implemented as, e.g., one or more computer programs on a respective computing device. The server (“server system”) 120 can be implemented as one or more computer programs on one or computing devices and is generally remote from the clients 110. That is, the server 120 and the clients 110 communicate with one another over a data communications network, e.g., the Internet. As part of the operation of the clients 110 and the server 120, the clients 110 transmit data (“messages”) 130 to the server 120 for use by the server 120. For example, the server 120 can collect analytics data, e.g., error reports, performance data, or other data, from the clients 110. The analytics data can be information about the operation of software running on the clients 110 or otherwise being monitored by the clients 110. Attorney Docket No.56113-0414WO1 As another example, the server 120 can collect training data from the respective clients 110 for use in training or re-training one or more machine learning models. In a conventional scheme, when the clients 110 submit data to the server 120, the server 120 can by default determine which client 110 each submission is received from. This can increase the information visible to the server 120 and therefore poses a potential data security and client privacy risk, e.g., because the server 120 can associate particular messages with particular clients 110. To address this issue, the system 100 modifies the communications between the server 120 and the clients 110 by causing some or all of the clients 110 to shuffle and re- encrypt ciphertexts 140 generated from the messages 130. In other words, the system 100 implements a single server, multi-client message shuffling scheme that allows the server 120 access to the messages 130 transmitted by the clients 110 without allowing the server 120 to identify which message 130 was transmitted by which client 110. At a high level, as part of the single server, multi-client message shuffling scheme, the server 120 obtains ciphertexts 140 from some or all of the clients 110. Each ciphertext 140 has been generated by encrypting a corresponding message 130, e.g., each ciphertext is an encrypted message or an encrypted message portion. The system 100 then performs multiple shuffling iterations that obscure the identity of which client 110 originally provided which ciphertext 140 to the server 120. At each of the shuffling iterations, one of the clients 110 shuffles and re-encrypts some or all of the ciphertexts 140. Performing shuffling iterations is described in more detail below. After performing the shuffling iterations, the system 100 provides the plurality of ciphertexts 140 for decryption. For example, the server 120 and the clients 110 can perform a distributed decryption that results in the server 120 possessing the decrypted messages 130 (also referred to as “plaintexts”). Examples of this are described below. Thus, while, as a result of performing the scheme, the server 120 gains access to the decrypted messages that are generated by decrypting the ciphertexts 140, the server 120 does not gain access to information that identifies which client 110 originally provided any given one of the decrypted messages. Additionally, there can be a large number of clients 110, but any given client 110 may be resource-constrained, have limited network connectivity, or both. Thus, computations assigned to a given client may “drop out” (fail) at any given time during the Attorney Docket No.56113-0414WO1 single server, multi-client message shuffling scheme. For example, the client 110 that is performing any given shuffling iteration may not be able to complete the shuffling iteration or may return an output that is not able to validated. Therefore, the system 100 implements the scheme to be robust to a certain number of client “drop outs.” Moreover, as described above, one or more of the clients 110 may be malicious actors that are attempting to (i) recover unencrypted versions of messages 130 from other clients or (ii) cooperate with the server 120 to allow the server 120 to identify the source of some or all of the messages 130 once the messages have been decrypted. Thus, the system 100 implements the scheme so that (i) the messages 130 are secure until decrypted on the server 120 even though ciphertexts are transmitted to various clients 110 and (ii) the source of each message 120 is obfuscated from the server 120 even if the server 120, potentially in collaboration with certain ones of the clients 110, attempts to discover the source of any given message 130. FIG.2A is a flow diagram of an example process 200 for transmitting messages between clients and a server. For convenience, the process 200 will be described as being performed by a system of one or more computers located in one or more locations. For example, a server system that is part of a message processing system, e.g., the message processing system 100 of FIG.1, appropriately programmed, can perform the process 200. The server system obtains a plurality of ciphertexts from a plurality of clients (step 202). That is, the server system obtains at least one ciphertext from each client in at least a subset, e.g., a proper subset, of the plurality of clients. For example, the subset of clients can be a proper subset and not all of the clients provide ciphertexts to the server. As another example, all of the clients can provide one or more ciphertexts to the server. For example, the server system can cause each of at least a subset of the plurality of clients to encrypt an original message to generate one or more ciphertexts from the original message. As a particular example of this, the server system can cause each of at least a subset of the plurality of clients to generate threshold shares of a private key for an encryption system and send a corresponding public key to the server system. As a particular example, the threshold shares can be Shamir shares, e.g., of a sampled secret key, and the encryption system can be ElGamal encryption. One example of the key generation process is described with reference to FIG.2B. Attorney Docket No.56113-0414WO1 FIG.2B is a flow diagram of an example process 250 for performing key generation. For convenience, the process 250 will be described as being performed by a system of one or more computers located in one or more locations. For example, a message processing system, e.g., the message processing system 100 of FIG.1, appropriately programmed, can perform the process 250. The server system S generates a partition that divides the clients into m committees ^^^through ^^^(step 252) with each committee having size ^^ௗ^^. Let ^^^,^be the jth member of ^^^. The result of the key generation process is then that ^^^,^receives a t-out-of- ^^ௗ^^Shamir share ^^ ^^^,^of a uniformly sampled secret key ^^ ^^ ∈ ℤ^, where t is a threshold parameter received as input by the system that defines the degree to which the secret key is recoverable. That, the secret key is only recoverable if t clients in the same committee reveal their shares. The server system S receives a public key ^^^^ ^^and ^ ^^^^ ^^ ^^, ^^^^∈^^^,^∈^^^^^^. the key generation process, each client ^^^,^samples a private key ^^^,^← ^^^and generates two independent t-out-of- ^^ௗ^^Shamir sharings of ^^^,^(step 254). The client can do this using polynomials ^^^,^≔ ^^^,^^ ∑௧^ୀ^^^^^^^, ^^^,^≔ ^^^,^^ ∑௧^ୀ^^^^^^^along with coefficient commitments ^^^,^,^≔ ^^^ೖ ^^∈^௧^, ^^^,^,ଶ≔ ^^ೖ ^^∈^௧^, ^^^,ೕfor verification. The two resulting sets ൫ ^^^ ^^^൯^∈^^^^^^and ^^^,^,ଶ≔ ൫ ^^^ ^^^൯^∈^^^^^^. ^^^,^,^and ^^^,^,ଶamongst ^^^and ^^^, respectively (step 256). Thus, clients in ^^^only need to generate one sharing. This communication can be performed via the server system S using deterministic public key encryption. The server system S stores the associated ciphertexts for later use (in step 262). Each client ^^^,^sends commitments ^^^,^,^, ^^^,^,ଶ, and ^^^^,ೕto the server system S (step 258) and S derives commitments to each expected share in ^^^,^,^and ^^^,^,ଶand sends them to the party receiving the corresponding share. The server system S stores commitments for later use (in steps 262 and 272). Each client ^^^,^checks their shares against the commitments and sends the (possibly empty) set of faulty shares to the server system S (step 260). Attorney Docket No.56113-0414WO1 The server system S confirms the discrepancies (or refutes them), drops the faulty clients and informs their neighbors, i.e., clients with whom faulty clients shared shares (step 262). For i > 1, each client ^^^,^then sums the shares it receives from (non-faulty and non-dropped out members of) ^^^to form ^^̃^,^and from ^^^ି^to form ^^^,^(step 264). Each client ^^^,^then sends offset^,^≔ ^^̃^,^െ ^^^,^to the server he server system S ^^^̃ T^,ೕfrom in ൫ ^^^,^,^൯^∈^and ^^^checks offset^,^against commitments in ^^^,^,^and ^^^,^,ଶ, bad shares (step 266). Let ^^^denote the value shared across committee i in the ^^̃^,^. For each i > 1, the server system recovers ^^^െ ^^^ି^, then computes ^^^≔ ^^^െ ^^^sends this value to each client in committee i (step 268). Each client ^^^,^constructs their Shamir ^^^,^ൌ ^^̃^,^െ ^^^(step 270). Thus, the distributed private key is ^^ ^^ ≔ ^^^. The server system S constructs its outputs ^^^^ ^^and ^ ^^^^ ^^ ^^, ^^^^∈^^^,^∈^^^^^^using the coefficient commitments (step 272). Optionally, once the server system receives the corresponding public keys, the server system can then mask each corresponding public key with the secret key to generate a masked key and then send the masked keys to the clients. The system can then cause each of the one or more clients to encrypt their corresponding message(s) with the masked keys to generate the plurality of ciphertexts. The server system can “cause” a client to perform encryption or another operation by sending a request to the client to perform the operation, by sending instructions that when executed by the client cause the client to perform the operation, or in any other appropriate manner. Returning to the description of FIG.2A, the system then performs multiple shuffling iterations (step 204). At each of the multiple shuffling iterations, one of the clients shuffles and re-encrypts some or all of the ciphertexts. The system can determine that a given shuffling iteration is the last iteration in any of a variety of ways. For example, at the last iteration, the system can determine that a threshold number of iterations have been performed and, in response, refrain from performing further iterations. Attorney Docket No.56113-0414WO1 As another example, as described above, some of the clients can “drop out” while the shuffling iterations are performed, e.g., due to network connectivity or insufficient resources that cause a given shuffling iteration to time out, or due to failing to provide a valid decryption. To account for this, the system can only terminate the shuffling iterations in response determining that a first threshold number of iterations have been performed and that less than a second threshold number of iterations have been unsuccessfully performed. For example, the second threshold number can be a predetermined ratio of the total number of clients. If the first threshold number of iterations have been performed and at least the second threshold number of iterations have been performed, the system can optionally determine that the shuffling scheme has failed and refrain from giving the server access to the decrypted messages. After performing the shuffling iterations, the system provides the plurality of ciphertexts for decryption (step 206). For example, the server system can cause the clients to perform a distributed decryption that results in the server system possessing the decrypted messages and without revealing the decrypted messages to any of the clients. An example technique for performing the decryption using the keys generated by performing the technique of FIG.2B is described in FIG.2C. FIG.2C is a flow diagram of an example process 280 for performing shared key decryption. For convenience, the process 280 will be described as being performed by a system of one or more computers located in one or more locations. For example, a message processing system, e.g., the message processing system 100 of FIG.1, appropriately programmed, can perform the process 280. At the outset of the process 280, the server system holds k ciphertexts ^ ^^^ൌ ൫ ^^^ℎ^^^ ^^, ℎ^൯^, where h is the group element and ^^^is the i-th message with ℎ^, ^^^∈ ^^. After the process 280 is completed, the server system receives the k messages ^^^. The server system sends group elements ^ℎ^^^∈^^^to each client . Each client ^^^,^in the a-th committee performs a series of operations on the received group elements (step 284). In particular, client ^^^,^: (a) computes ^^ ൌ ℎ^^ ^^ ^^, ^^^ (b) ∈ ℤ^ Attorney Docket No.56113-0414WO1 (c) computes ^^ ൌ ^^^ೕand fo^ೕ^ r each I, ^^^,^ൌ ℎ^, and (d) sends ( ^^^^ , ^^ ^^^,^, ^^^,^^^ୀ^) to For each j, the server system S a ^^^← ℤ^to ^^^,^(step 286). The client ^^^,^sends ^^^ൌ ^^^^ ^^ ^^ ^^^,^to the server system S (step 288). For each j, the server system S performs checks (a) and (b), i.e., checks whether: (a) ^^^^^௨ೕ ^^ ^^ೌ,ೕൌ ^^^ೕand ^ೕhold, for all i (step 290). (b) pass for at least t clients, then server system S recovers ℎ^^^ ^^ (via in the exponent) and ^^^(via group division), for each i (step 292). FIG.3A is a flow diagram of an example process 300 for performing a shuffling iteration. For convenience, the process 300 will be described as being performed by a system of one or more computers located in one or more locations. For example, a message processing system, e.g., the message processing system 100 of FIG.1, appropriately programmed, can perform the process 300. The server system provides a respective subset of the ciphertexts to a respective client (step 302). That is, the system identifies a subset of the ciphertexts for the shuffling iteration and a client for the shuffling iteration. The system then provides the identified subset of ciphertexts to the identified client. For example, the system can transmit the identified subset of ciphertexts to the identified client. As one example, the system can select, e.g., randomly or according to a predetermined mapping of clients to shuffling iterations, a client for the shuffling iteration and then provide all of the ciphertexts to the selected client. As another example, the system can provide only a proper subset of the ciphertexts to the selected client. As a particular example of this, the server system can maintain data representing the plurality of ciphertexts as a two-dimensional grid of ciphertexts. In this example, at each iteration, the respective subset can include only the ciphertexts in a corresponding row or a corresponding column of the two-dimensional grid. One example of such a scheme is described in more detail below with reference to FIGS.4A and 4B. Attorney Docket No.56113-0414WO1 The server system obtains, from the respective client, shuffled and re-encrypted ciphertexts (step 304). That is, the server system causes the respective client to shuffle, i.e., randomly permute, and re-encrypt (or re-encrypt and shuffle) the subset of ciphertexts. Thus, the shuffled and re-encrypted ciphertexts include the same number of ciphertexts as in the respective subset. The server system updates the respective subset of the ciphertexts to be the shuffled and re-encrypted ciphertexts (step 306). In some implementations, the server system also obtains, from the respective client, a proof of the validity of the shuffled and re-encrypted ciphertexts. For example, the proof of the validity can be a zero-knowledge proof computing using the protocol described in S. Bayer and J. Groth. Efficient zero-knowledge argument for correctness of a shuffle. In D. Pointcheval and T. Johansson, editors, Advances in Cryptology – EUROCRYPT 2012, pages 263–280, Berlin, Heidelberg, 2012. Springer Berlin Heidelberg. ISBN 978-3-642-29011-4. The proof is referred to as “zero-knowledge” because determining the proof does not require decrypting any of the ciphertexts. In these implementations, the system updates the respective subset of the ciphertexts to be the shuffled and re-encrypted ciphertexts only in response to verifying the proof of the validity. If the server system cannot verify the proof, the system leaves the respective subset of the ciphertexts unchanged, i.e., does not update the respective subset. Depending on how the system assigns subsets to clients and how the system performs shuffling iterations, the system can implement any of a variety of shuffling schemes. One example of a shuffling scheme in which the server sends each of the ciphertexts to the client at every shuffling iteration is described in FIG.3B. FIG.3B is a flow diagram of an example process 350 for carrying out a shuffling scheme. For convenience, the process 350 will be described as being performed by a system of one or more computers located in one or more locations. For example, a message processing system, e.g., the message processing system 100 of FIG.1, appropriately programmed, can perform the process 350. At the beginning of the shuffling scheme, the server system holds a set of ciphertexts and, at the end of the shuffling scheme, the server holds a shuffled re- encryption of the ciphertexts. Attorney Docket No.56113-0414WO1 Prior to performing the shuffling scheme, the system places the clients in a random order, so that L is a collection of ^^^^௨^clients in a random order chosen by the server system S. The system then performs a respective shuffling iteration for each client according to the random order. At any given shuffling iteration, the server system sends the current set of ciphertexts to the client sc for the shuffling iteration (step 352). The client sc for the shuffling iteration shuffles and re-encrypts the current set of ciphertexts (step 354). In particular, the client sets ^^^^ ^^≔ ^^^^ ^^൫re െ encrypt^ ^^^௨^^^^௧^൯, where π is a uniformly random permutation and ^^^௨^^^^௧is the current set of ciphertexts. The client sc for the shuffling iteration constructs a proof P that it has shuffled and re-encrypted the current set of ciphertexts (step 356). The client sc for the shuffling iteration sends the proof P and the shuffled and re- encrypted set of ciphertexts ^^^^ ^^to the server (step 358). If the server receives the proof P and the shuffled and re-encrypted set of ciphertexts ^^^^ ^^successfully and the proof P is valid, the server sets ^^^௨^^^^௧≔ ^^^^ ^^(step 360). Otherwise, it leaves ^^^௨^^^^௧unchanged and drops sc. If ^^^^௨^െ ^^ clients have provided valid shuffles, the system terminates the shuffling scheme. Here, d is the dropout limit, i.e., the maximum number of clients that can drop out before the shuffling scheme is invalid. On the other hand, if d + 1 clients have dropped out, the system aborts the shuffling scheme. Another example of a shuffling scheme, referred to as an amortized shuffler scheme, is described in FIG.3C. FIG.3C is a flow diagram of an example process 370 for carrying out an amortized shuffler scheme. For convenience, the process 370 will be described as being performed by a system of one or more computers located in one or more locations. For example, a message processing system, e.g., the message processing system 100 of FIG. 1, appropriately programmed, can perform the process 370. The system performs distributed key generation (step 372). During distributed key generation, the server system S partitions the clients into m committees of size ^^ௗ^^uniformly at random. All parties then perform the key generation described above with reference to FIG.2B. The server system S generates sk′ at random, Attorney Docket No.56113-0414WO1 sets sk′′ = sk + sk′ and sends the resulting key ^^^^ ^^ᇲᇲൌ ^^^^ ^^^^^^ ^^ᇲto all clients to use as a public key. Each client c then computes the ciphertext൫^^^^^^^ ^^ᇲᇲ^, ^^^^൯and sends it to the server. The system performs distributed verifiable shuffling (step 374). In particular, the system performs the shuffling scheme described above with reference to FIG.3B with ^^^^௨^randomly selected clients and dropout limit d, all the ciphertexts from all the clients, and the new public key pk’= ^^^^ ^^ା௧. That is, the shuffler, i.e., the client that performs the shuffling iteration, also adds an offset to the public key that is used. This is done so as to protect against another possible adversary who controls more than a specified fraction of clients. In the amortized shuffler scheme, the server system randomly selects a subset of the clients to perform the sequence of shuffling iterations, thereby amortizing the computation cost of shuffling and re-encrypting the ciphertexts. The system performs distributed decryption (step 376). During distributed decryption, the server system uses key homomorphism to return the key to sk from sk’’. The server system also partitions the ciphertexts resulting from performing step 374 into m groups and for each group ^^^runs the process of FIG. 2C with ^^^and ^^^. The server system then takes the resulting plaintexts, i.e., decrypted messages, as output. Another example of a shuffling scheme, referred to as an alternating shuffler scheme, is described below with reference to FIG.4A and FIG.4B. FIG.4A shows an example 400 of performing an alternating shuffler scheme. To perform this scheme, the server system receives n ciphertexts 410. The server system then generates a grid 430 by shuffling the ciphertexts using a public permutation and then rearranging the shuffled ciphertexts 420 into an h x w grid. The system then performs l private iterations 440, where l is a fixed integer greater than one, e.g., 2 or 3. At each private iteration 440, the system shuffles each row of the grid privately, i.e., by performing one or more shuffling iterations, and then transposes the grid. As shown in FIG.4A, the system makes use of the h x w grid to perform shuffling iterations. Attorney Docket No.56113-0414WO1 In particular, at each of the l update iterations, the system performs, in parallel, a sequence of shuffling iterations for each of the h rows in the grid. The result of the sequence shuffling iterations for a given row is an updated row of ciphertexts. For example, each sequence of shuffling iterations can be assigned to a different subset of the clients, with each shuffling iteration within the sequence being performed by a different, randomly selected client within the subset. Thus, each row of the grid is updated at each of the l iterations. At each of the l update iterations, once the rows of the grid have been updated, the system permutes the grid. Because of the way the alternating shuffler scheme is structured, the alternating shuffling scheme is cryptographically secure in the honest-but-curious single-server threat model, uses sublinear communication per user, requires a small number of rounds, and is resilient to both dropouts and a small fraction of users colluding with the server. The alternating shuffler scheme is described further below with reference to FIG. 4B. FIG.4B is a flow diagram of an example process 470 for carrying out an alternating shuffler scheme. For convenience, the process 470 will be described as being performed by a system of one or more computers located in one or more locations. For example, a message processing system, e.g., the message processing system 100 of FIG. 1, appropriately programmed, can perform the process 470. The system performs distributed key generation (step 472). During distributed key generation, the server system S partitions the clients into m committees of size ^^ௗ^^uniformly at random. All parties then perform the key generation described above with reference to FIG.2B and the server system sends the resulting key ^^^^ ^^to all clients to use as a public key. Each client c then computes a ciphertext and sends it to the server system. The server system assigns the ciphertexts into an h x w grid M according to a permutation π. The system performs distributed verifiable shuffling using the grid M (step 474). The server system generates t at random and changes all ciphertexts to be encrypted with sk′ = sk + t using key homomorphism and distributes the new public key ^^^^ ^^ା௧. Attorney Docket No.56113-0414WO1 The server system splits the clients into shuffling committees of size ^^^^௨^. The system then repeats the following operations l times: (a) The server system S conducts h parallel instances of the protocol of Figure 3B (one per row in M). Each instance is run with one of the shuffling committees, e.g., chosen to spread the load as evenly as possible over those committees, and inputs sk′ and ciphertexts in a row in M. (b) The server system S replaces each row of ciphertexts in M with its shuffled version. (c) The server system S transposes M. At the conclusion of the l iterations, the server system S has a grid of re-encrypted and shuffled ciphertexts. The system performs distributed decryption (step 476). During distributed decryption, the server system uses key homomorphism to return the key to sk from sk’. The server system also partitions the ciphertexts resulting from performing step 374 into m groups and for each group ^^^runs the process of FIG.2C with ^^^and ^^^. The server system then takes the resulting plaintexts, i.e., decrypted messages, as output. This specification uses the term “configured” in connection with systems and computer program components. For a system of one or more computers to be configured to perform particular operations or actions means that the system has installed on it software, firmware, hardware, or a combination of them that in operation cause the system to perform the operations or actions. For one or more computer programs to be configured to perform particular operations or actions means that the one or more programs include instructions that, when executed by data processing apparatus, cause the apparatus to perform the operations or actions. Embodiments of the subject matter and the functional operations described in this specification can be implemented in digital electronic circuitry, in tangibly-embodied computer software or firmware, in computer hardware, including the structures disclosed in this specification and their structural equivalents, or in combinations of one or more of them. Embodiments of the subject matter described in this specification can be implemented as one or more computer programs, i.e., one or more modules of computer program instructions encoded on a tangible non transitory storage medium for execution by, or to control the operation of, data processing apparatus. The computer storage Attorney Docket No.56113-0414WO1 medium can be a machine-readable storage device, a machine-readable storage substrate, a random or serial access memory device, or a combination of one or more of them. Alternatively or in addition, the program instructions can be encoded on an artificially generated propagated signal, e.g., a machine-generated electrical, optical, or electromagnetic signal, that is generated to encode information for transmission to suitable receiver apparatus for execution by a data processing apparatus. The term “data processing apparatus” refers to data processing hardware and encompasses all kinds of apparatus, devices, and machines for processing data, including by way of example a programmable processor, a computer, or multiple processors or computers. The apparatus can also be, or further include, special purpose logic circuitry, e.g., an FPGA (field programmable gate array) or an ASIC (application specific integrated circuit). The apparatus can optionally include, in addition to hardware, code that creates an execution environment for computer programs, e.g., code that constitutes processor firmware, a protocol stack, a database management system, an operating system, or a combination of one or more of them. A computer program, which may also be referred to or described as a program, software, a software application, an app, a module, a software module, a script, or code, can be written in any form of programming language, including compiled or interpreted languages, or declarative or procedural languages; and it can be deployed in any form, including as a stand alone program or as a module, component, subroutine, or other unit suitable for use in a computing environment. A program may, but need not, correspond to a file in a file system. A program can be stored in a portion of a file that holds other programs or data, e.g., one or more scripts stored in a markup language document, in a single file dedicated to the program in question, or in multiple coordinated files, e.g., files that store one or more modules, sub programs, or portions of code. A computer program can be deployed to be executed on one computer or on multiple computers that are located at one site or distributed across multiple sites and interconnected by a data communication network. In this specification, the term “database” is used broadly to refer to any collection of data: the data does not need to be structured in any particular way, or structured at all, and it can be stored on storage devices in one or more locations. Thus, for example, the index database can include multiple collections of data, each of which may be organized and accessed differently. Attorney Docket No.56113-0414WO1 Similarly, in this specification the term “engine” is used broadly to refer to a software-based system, subsystem, or process that is programmed to perform one or more specific functions. Generally, an engine will be implemented as one or more software modules or components, installed on one or more computers in one or more locations. In some cases, one or more computers will be dedicated to a particular engine; in other cases, multiple engines can be installed and running on the same computer or computers. The processes and logic flows described in this specification can be performed by one or more programmable computers executing one or more computer programs to perform functions by operating on input data and generating output. The processes and logic flows can also be performed by special purpose logic circuitry, e.g., an FPGA or an ASIC, or by a combination of special purpose logic circuitry and one or more programmed computers. Computers suitable for the execution of a computer program can be based on general or special purpose microprocessors or both, or any other kind of central processing unit. Generally, a central processing unit will receive instructions and data from a read only memory or a random access memory or both. The essential elements of a computer are a central processing unit for performing or executing instructions and one or more memory devices for storing instructions and data. The central processing unit and the memory can be supplemented by, or incorporated in, special purpose logic circuitry. Generally, a computer will also include, or be operatively coupled to receive data from or transfer data to, or both, one or more mass storage devices for storing data, e.g., magnetic, magneto optical disks, or optical disks. However, a computer need not have such devices. Moreover, a computer can be embedded in another device, e.g., a mobile telephone, a personal digital assistant (PDA), a mobile audio or video player, a game console, a Global Positioning System (GPS) receiver, or a portable storage device, e.g., a universal serial bus (USB) flash drive, to name just a few. Computer readable media suitable for storing computer program instructions and data include all forms of non volatile memory, media and memory devices, including by way of example semiconductor memory devices, e.g., EPROM, EEPROM, and flash memory devices; magnetic disks, e.g., internal hard disks or removable disks; magneto optical disks; and CD ROM and DVD-ROM disks. To provide for interaction with a user, embodiments of the subject matter described in this specification can be implemented on a computer having a display device, e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor, for displaying Attorney Docket No.56113-0414WO1 information to the user and a keyboard and a pointing device, e.g., a mouse or a trackball, by which the user can provide input to the computer. Other kinds of devices can be used to provide for interaction with a user as well; for example, feedback provided to the user can be any form of sensory feedback, e.g., visual feedback, auditory feedback, or tactile feedback; and input from the user can be received in any form, including acoustic, speech, or tactile input. In addition, a computer can interact with a user by sending documents to and receiving documents from a device that is used by the user; for example, by sending web pages to a web browser on a user’s device in response to requests received from the web browser. Also, a computer can interact with a user by sending text messages or other forms of message to a personal device, e.g., a smartphone that is running a messaging application, and receiving responsive messages from the user in return. Data processing apparatus for implementing machine learning models can also include, for example, special-purpose hardware accelerator units for processing common and compute-intensive parts of machine learning training or production, i.e., inference, workloads. Machine learning models can be implemented and deployed using a machine learning framework, e.g., a TensorFlow framework or a Jax framework. Embodiments of the subject matter described in this specification can be implemented in a computing system that includes a back end component, e.g., as a data server, or that includes a middleware component, e.g., an application server, or that includes a front end component, e.g., a client computer having a graphical user interface, a web browser, or an app through which a user can interact with an implementation of the subject matter described in this specification, or any combination of one or more such back end, middleware, or front end components. The components of the system can be interconnected by any form or medium of digital data communication, e.g., a communication network. Examples of communication networks include a local area network (LAN) and a wide area network (WAN), e.g., the Internet. The computing system can include clients and servers. A client and server are generally remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other. In some embodiments, a server transmits data, e.g., an HTML page, to a user device, e.g., for purposes of displaying data to and receiving user input from a user interacting with Attorney Docket No.56113-0414WO1 the device, which acts as a client. Data generated at the user device, e.g., a result of the user interaction, can be received at the server from the device. While this specification contains many specific implementation details, these should not be construed as limitations on the scope of any invention or on the scope of what may be claimed, but rather as descriptions of features that may be specific to particular embodiments of particular inventions. Certain features that are described in this specification in the context of separate embodiments can also be implemented in combination in a single embodiment. Conversely, various features that are described in the context of a single embodiment can also be implemented in multiple embodiments separately or in any suitable subcombination. Moreover, although features may be described above as acting in certain combinations and even initially be claimed as such, one or more features from a claimed combination can in some cases be excised from the combination, and the claimed combination may be directed to a subcombination or variation of a subcombination. Similarly, while operations are depicted in the drawings and recited in the claims in a particular order, this should not be understood as requiring that such operations be performed in the particular order shown or in sequential order, or that all illustrated operations be performed, to achieve desirable results. In certain circumstances, multitasking and parallel processing may be advantageous. Moreover, the separation of various system modules and components in the embodiments described above should not be understood as requiring such separation in all embodiments, and it should be understood that the described program components and systems can generally be integrated together in a single software product or packaged into multiple software products. Particular embodiments of the subject matter have been described. Other embodiments are within the scope of the following claims. For example, the actions recited in the claims can be performed in a different order and still achieve desirable results. As one example, the processes depicted in the accompanying figures do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some cases, multitasking and parallel processing may be advantageous. What is claimed is:
Claims
Attorney Docket No.56113-0414WO1 CLAIMS 1. A method comprising: obtaining, by a server computer system, a plurality of ciphertexts from a plurality of clients; at each of a plurality of iterations: providing a respective subset of the ciphertexts to a respective client; obtaining, from the respective client, shuffled and re-encrypted ciphertexts; and updating the respective subset of the ciphertexts to be the shuffled and re- encrypted ciphertexts; and after the last iteration of the plurality of iterations, providing the plurality of ciphertexts for decryption.
2. The method of claim 1, wherein providing the plurality of ciphertexts for decryption comprises: providing the plurality of ciphertexts to the plurality of clients for decryption.
3. The method of claim 2, further comprising: after providing the plurality of ciphertexts to the plurality of clients for decryption, obtaining decrypted ciphertexts from the plurality of clients 4. The method of any preceding claim, wherein obtaining, by a server computer system, a plurality of ciphertexts from a plurality of clients comprises: causing each of at least a subset of the plurality of clients to encrypt an original message to generate one or more ciphertexts from the original message.
5. The method of claim 4, wherein causing each of at least a subset of the plurality of clients to encrypt an original message to generate one or more ciphertexts from the original message comprises: causing each of at least a subset of the plurality of clients to generate threshold shares of a private key for an encryption system and send a corresponding public key to the server system.
6. The method of claim 5, wherein the threshold shares are Shamir shares.Attorney Docket No.56113-0414WO1 7. The method of claim 5 or claim 6, wherein the encryption system is ElGamal encryption.
8. The method of any one of claims 5-7, wherein causing each of at least a subset of the plurality of clients to encrypt an original message to generate one or more ciphertexts from the original message further comprises: masking each corresponding public key with a secret key to generate a masked key; sending the masked keys to the clients; and causing the clients to encrypt the messages with the masked keys to generate the plurality of ciphertexts.
9. The method of any preceding claim, wherein providing a respective subset of the ciphertexts to a respective client comprises: providing all of the plurality of ciphertexts to the respective client.
10. The method of any one of claims 1-8, wherein the server system maintains data representing the plurality of ciphertexts as a two-dimensional grid of ciphertexts, and wherein, at each iteration, the respective subset includes only the ciphertexts in a corresponding row or a corresponding column of the two-dimensional grid.
11. The method of any preceding claim, wherein obtaining, from the respective client, shuffled and re-encrypted ciphertexts comprises: obtaining, from the respective client, a proof of the validity of the shuffled and re- encrypted ciphertexts; and wherein updating the respective subset of the ciphertexts to be the shuffled and re- encrypted ciphertexts comprises: updating the respective subset of the ciphertexts to be the shuffled and re- encrypted ciphertexts only in response to verifying the proof of the validity.
12. The method of any preceding claim, further comprising: at the last iteration: determining that a threshold number of iterations have been performed and, in response, refraining from performing further iterations.Attorney Docket No.56113-0414WO1 13. The method of any preceding claim, further comprising: at the last iteration: determining that a first threshold number of iterations have been performed and less than a second threshold number of iterations have been unsuccessfully performed and, in response, refraining from performing further iterations.
14. A system comprising one or more computers and one or more storage devices storing instructions that when executed by the one or more computers cause the one more computers to perform the operations of the respective method of any one of claims 1-13.
15. One or more computer storage media storing instructions that when executed by one or more computers cause the one more computers to perform the operations of the respective method of any one of claims 1-13.
Citation Information
Patent Citations
Encrypted Search Cloud Service with Cryptographic Sharing
US20180314847A1
Oblivious outsourcing of file storage
WO2018050911A1
